Plurals in SLI: Prosodic Deficit or Morphological Deficit?

Abstract

Two accounts for the segmental and prosodic anomalies observed in plurals pro duced by 5 adults with specific language impairment (SLI), one prosodic and one morphological, are compared. The prosodic account proposed is that the grammars of these individuals do not tolerate extraprosodicity: indirect licensing by the pro sodic word. Although this account can capture a range of facts, it is rejected for several reasons, the most significant of which is that it cannot discriminate between anomalous outputs such as [dDg.s] and natural-sounding outputs such as [dngz], both of which are produced by impaired speakers. In view of this, a morphological account is proposed: The grammars of these impaired individuals lack certain sublexical features, in particular [?plural]; the notion of plurality is expressed at the level of conceptual structure. Consequently, plurals must be built through com pensatory means. They may involve the concatenation of stems and thereby struc turally resemble compounds, both morphologically and prosodically (yields [dog.s]). They may be stored as morphologically unanalyzed chunks (yields [dDgz]). Evidence in support of both options is provided.